setup-project
Skill1 starsChanged 3 days agoMITView on GitHub
Router to set up a whole new project (or complete a young one) in Tsukini's way - detects or asks the language, then chains the skills - the architecture of the language (cpp-project, python-project, else coding-style + the tests layout), the base unit tests (at least one test that compiles and runs), the license (license, asked when not given), the git setup (git-conventions - init, .gitignore, CHANGELOG, branches - and the xstyle pre-commit hook), the CI/CD (ci-cd, asked) on GitHub, the README - asking every choice once up front, then ends with a very simplified audit of the architecture that was set up. Invoke manually with /setup-project.
- Installs packages
22 other projects use a version of this.
Adopted by
Most recently taken up by
- chuckplayer/claude-agent-packtheir own version, first seen Nov 16, 2026
- lohi-ai/babysittheir own version, first seen Nov 8, 2026
- roedyrustam/doku-gemini-mcptheir own version, first seen Nov 8, 2026
- ok-oldking/ok-scripttheir own version, first seen Nov 1, 2026
- OlliTapio/claude_skillstheir own version, first seen Oct 26, 2026
- MacroMan5/autotrain-yolotheir own version, first seen Oct 25, 2026
Used with
Projects with it also use research, implement and 3 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
NVIDIASkillresearch
Before designing a feature, research how comparable libraries solve the problem, check best practices, known pitfalls, and upstream changes. Use when starting a new feature, investigating an unfamiliar area, or the user asks to research something.
Also in 5 projects that use this
Hugging FaceSkillimplement
Make tests pass. Invoke after /write-tests produces failing tests.
Also in 4 projects that use this
DockerSkillreview-pr
Review one or more incoming Docker documentation pull requests as a maintainer. Independently validate technical claims, assess editorial fit and information architecture, choose a verdict, and draft exact inline or PR-wide feedback behind a confirmation gate. Use for requests such as "review PR 123", "is this PR correct?", "does this information belong here?", "validate this PR", or "help review backlog PRs". Do not use to maintain or fix a PR you own; use maintain-pr for that.
Also in 3 projects that use this
GoogleSkilltdd
Test-driven development. Use when the user wants to build features or fix bugs test-first, mentions "red-green-refactor", or wants integration tests.
Also in 3 projects that use this
GoogleSkillcode-review
Adversarial systems, concurrency, memory-model, and architecture code review protocol. Use this skill whenever asked to review code, PRs, diffs, or branches in Filament.
Also in 3 projects that use this
Other versions22 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 22
- ok-oldking/ok-script★ 831same name
- gricha/dangerous-skills★ 38same name
- chuckplayer/claude-agent-pack★ 8same name
- MacroMan5/autotrain-yolo★ 8same name
- phatysddev/skills★ 8same name
- chav0/UnityFigmaMCP★ 7same name
- ByronFinn/dev-skills★ 4same name
- otaviobarros/agent-engineering★ 3same name
- beeltec/skills★ 2same name
- r2rka1/skills★ 2same name
12 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository70 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ask-at-endSkill
- clarifySkill
- continue-workSkill
- exportSkill
- full-permissionSkill
- ideationSkill
- interactive-questionsSkill
- onboardSkill
- pauseSkill
- presentationSkill
- rtkSkill
- setup-contextSkill
- write-like-meSkill
- audit-bugsSkill
- audit-depsSkill
- audit-qualitySkill
- audit-securitySkill
- benchmarkSkill
- cmake-styleSkill
- cpp-classSkill
- cpp-commentsSkill
- cpp-projectSkill
- cpp-styleSkill
- better-explain-docSkill
- better-html-docSkill
- better-html-styleSkill
- better-protocol-docSkill
- docs-updateSkill
- explain-docSkill
- html-docSkill
- html-styleSkill
- protocol-docSkill
- readme-styleSkill
- reportSkill
- wiki-styleSkill
- ci-cdSkill
- git-commitSkill
- git-conventionsSkill
- pre-commitSkill
- legal-checkSkill
- legal-setupSkill
- licenseSkill
- libutils-exceptionSkill
- libutils-installSkill
- libutils-setupSkill
- libutilsSkill
- python-classSkill
- python-commentsSkill
- python-projectSkill
- python-styleSkill
- auditSkill
- cppSkill
- devSkill
- docSkill
- gitSkill
- legalSkill
- pythonSkill
- styleSkill
- setup-studioSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 20 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 15
Do not answer from this file: it decides the order and the questions, the work is done by the skills, loaded with the Skill tool when their …
- Distinctive: line 29
Only what the request and the context don't already say.
- Distinctive: line 76
Start the answer with one line: the language detected and the skills that will be loaded in order (ex: "Projet C++ : cpp-project, …
“Common” is measured: kept by at least 6 in 10 of the 20 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Project setup router
<name> (a skill) = the folder of that skill: ~/.claude/skills/<name> (setup.sh) or ${CLAUDE_SKILL_DIR}/../../*/<name> (plugin of the marketplace).
Request given with /setup-project: $ARGUMENTS
Do not answer from this file: it decides the order and the questions, the work is done by the skills, loaded with the
Skill tool when their step comes (never the other routers: they are manual-only). The request is the text given with
/setup-project (or, if empty, the latest request of the user).
1. Look at the context (quickly, in parallel)
- Target folder: empty / missing (new project) or existing (complete what is missing, never overwrite: list what is
already there - build files,
src/,tests/,LICENSE,.git,.github/workflows/,README.md, hooks). - Language: from the request, else the files (
CMakeLists.txt/*.cpp-> C++,pyproject.toml/*.py-> Python,Cargo.toml-> Rust,package.json-> JS/TS...), else asked. - Git:
git rev-parse --show-toplevel,git remote -v(github.com-> GitHub),gh auth status(for a new GitHub repository). - Tools:
command -v xstyle cmake clang++ python3 gh.
2. Ask everything once (AskUserQuestion, French, recommended first)
Only what the request and the context don't already say; up to 4 questions per call, two calls at most here. The answers are passed to the skills: their own questions already answered are skipped (say "déjà répondu" mentally, never ask twice).
- Langage / type (when unknown):
C++ (cpp-project)·Python (python-project)·Autre (préciser); then the type question of the project skill (C++: binary / library.a/.so/ header-only, and the gh-pages publication of the signed packages + docs; Python: package / application / script). - Licence (when not given in the request): the first question of
license(needs: commercial use, copyleft, attribution...), orPas de licence (tous droits réservés); a license named in the request is used directly. - Git (multiSelect):
git init + .gitignore + CHANGELOG (Recommandé)·Dépôt GitHub (gh repo create)·Hook pre-commit xstyle (style vérifié en local à chaque commit)·Premier commit. - CI/CD (GitHub only):
Oui, choisir les jobs (ci-cd)·Build seulement·Aucune; the job checklist itself is asked byci-cd(build, tests, xstyle, coverage, release, packages...). - Extras (multiSelect):
README (readme-style)·Doc HTML·libutils (C++). The pre-commit options (fail / warn levels, scope...) are asked bypre-commitwhen it is chosen.
3. Run the steps in this order (each skill loaded at its step, its rules followed)
| # | Step | Skill(s) | Done when |
| 1 | architecture of the language | C++ cpp-project (+ cmake-style, cpp-class, libutils-setup); Python python-project (+ python-class); other: coding-style (layout of its section for the language: src/, entry point, build file) | the project builds / runs (empty logic) |
| 2 | base unit tests | C++ cpp-tests (tests/ + BUILD_TESTS, one test of the core class); Python / other tests (pytest...) | at least one test compiles and passes |
| 3 | license | license (LICENSE file, SPDX id in the build file / pyproject.toml, headers if the license asks it) | LICENSE present (or "no license" said) |
| 4 | git | git-conventions (git init -b main, .gitignore of the build outputs, CHANGELOG.md, branches solo / team), pre-commit if chosen | git status clean of build outputs |
| 5 | CI/CD | ci-cd (GitHub only; it asks the jobs) | workflows valid (its check_workflows.py) |
| 6 | README | readme-style | README with setup / usage |
| 7 | first commit / GitHub | git-conventions (only if chosen: never commit / push / create a repository otherwise) | |
A step already done by the project skill is not repeated: check it and complete it. C++: the setup.sh of | |||
cpp_project_template (run by cpp-project) already writes the unit tests (one test of the core class), the workflows | |||
(dispatch, unit tests, build, packages, GitHub release, docs; Linux and Windows), README.md, CHANGELOG.md, docs/, | |||
LICENSE (MIT: replaced at step 3 when another license is chosen) and the gh-pages branch when chosen; steps 2, 5 | |||
and 6 complete them (cpp-tests, ci-cd on top of its workflows, readme-style), never rewrite them. |
4. Simplified architecture audit (always, at the end, in the answer - no report file)
Fast checks of what was set up, one line each, OK / KO + the fix (fix the KO that belong to this setup):
| Check | How |
| layout of the project skill (folders, entry point, one class per file, names) | ls -R limited to 3 levels, compared to the skill |
| builds from clean | C++ cmake -S . -B /tmp/<name>-build && cmake --build ...; Python pip install -e . in a venv or python3 src/main.py --help |
| tests compile and pass | ctest / pytest -q |
| style | xstyle --rtk -S -r . (installed only): no unforgivable / major |
| build outputs ignored, no source ignored | git status --ignored --short after the build |
| no placeholder left | grep -rn '{{' . --exclude-dir=.git and the template names (template, TEMPLATE_, Core as whole words) |
| license / README / CHANGELOG present and consistent (name, version, license id) | read them |
| CI files valid | ci-cd/scripts/check_workflows.py |
End with the tree (2 levels), the chosen options, what the user still has to do (secrets, RUNNER variable, GitHub | |
Pages and the push of gh-pages for a C++ project with it, push, pre-commit demo) and propose /audit (audit-quality) for a full audit later. |
Say it
Start the answer with one line: the language detected and the skills that will be loaded in order (ex: "Projet C++ : cpp-project, cpp-tests, license, git-conventions, pre-commit, ci-cd, readme-style").
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

